The invention relates to determining the geolocation of wireless devices. Specifically, the invention relates to rapidly and autonomously estimating Doppler-shift and other receiver frequency errors in a ranging-receiver when receiving GPS satellite signals, including signals weakened by obstructions in the signal path.
In the application of GPS technology to the geolocation of portable wireless devices, it is desirable, because of the limited battery power available in such wireless devices, that the geolocation function: (i) operates with minimum energy consumption; and (ii) completes a geolocating fix in a minimum time.
At or near the surface of the earth, the signal level of GPS satellites is less than the thermal noise level of the conventional GPS receivers due to: (i) the spread spectrum modulation of the signal; (ii) the height of the 12 hour-period orbit (approximately 11,000 miles) of the GPS satellites; and (iii) the limited transmitting power of the satellite. Conventional GPS receivers must use sophisticated signal processing techniques, such as code correlation techniques, to extract the satellite signals.
Furthermore, wireless devices are typically operated in situations where views of the sky, and hence direct line-of-sight with GPS satellites, are frequently obstructed, resulting in further reduction of the levels of the received signals. Consequently, the obstructed signal levels from the GPS satellites are well below the threshold at which conventional GPS receivers can receive reliable position-tracking signals, and certainly below the level where any receiving equipment can receive reliable GPS satellite ephemeris data.
In order to recover these relatively weak satellite signals, the signals must be synchronously de-spread using convolution or correlation techniques. In conventional autonomous GPS receivers with unobstructed signals, this is done by searching for the signals with trial frequencies (since the Dxc3x6ppler-shifis are unknown) and with trial Course Acquisition Codes (C/AC), also referred to as the satellite""s xe2x80x9cGold Codesxe2x80x9d, (since the satellite identities are also unknown). Since the unobstructed signals have relatively good signal-to-noise ratios, a small sample of the signal is sufficient for the correlation search step, and Doppler-shift frequency trial steps can be quite large, so the search proceeds quickly, especially when multiple channels are searched simultaneously (rough estimate of search space of about 3000 frequency/code xe2x80x9cboxesxe2x80x9d of 5 milliseconds/box=15 seconds). However, when the signals are further weakened, typically by obstructions in the signal path, then many more samples are required (perhaps hundreds to thousands of times more) than are needed for the unobstructed case, and the Dxc3x6ppler-shift search frequency step must be much smaller because of the longer correlation samples. The search may then take a very much longer time (rough estimate of search space of about 300,000 frequency/code xe2x80x9cboxesxe2x80x9d of 1 second=a few days). What is needed is a faster and more direct method than the above correlation and frequency stepping search approach.
However, because of the repetitive nature of certain aspects of the GPS satellite signals, it is possible to use techniques such as very narrow-band filtering or signal averaging. In narrowband filtering it is possible to use the Fast Fourier Transform (xe2x80x9cFFTxe2x80x9d) to process long segments of the signal, thereby creating large numbers of narrow xe2x80x9cfrequency-binsxe2x80x9d. For example, since xe2x80x9cbinxe2x80x9d width is inversely proportional to segment length, for a segment length of 1 (one) second, the bin width would be 1 Hz and for a signal bandwidth of 2 MHz, the resulting number of bins would 2 (two) million. The energy of the noise present in the bin-bandwidth is approximately evenly spread across all the xe2x80x9cfrequency binsxe2x80x9d. However, the signal energy only falls in a very small number of the bin, thereby making it possible to be better detect the signal against this lowered noise background. In signal averaging, segments of the signal that repeats exactly are coherently added to each other to improve the relationship between the desired signal and the noise.
Both of these processes trade time for signal-to-noise ratio (e.g., either coherent signal energy accumulation or longer signals, enabling narrower analysis bandwidth), to improve the repetitive aspect of the signals against the random noise present in the signals. Both of these processes are possible due to the presence of the aforesaid repetitive properties in the satellite signals, such as the pseudo random ranging code, which repeats exactly every code cycle, or the carrier, which repeats continuously except for modulation changes. When such repetitive signal samples are added coherently, they accumulate linearly (i.e., proportionally to the number of samples). On the other hand, the random noise present in the signals is not coherent, and (in the coherent summation of the repetitive aspect above) accumulates as the square root of the number of samples. For example, with one thousand samples, the repetitive signal would be a thousand times larger, while the noise component would be only approximately thirty times larger.
Therefore, by utilizing specialized algorithms, it is possible to accumulate and process enough of the satellite""s signals to obtain reliable tracking information. Tracking information includes Dxc3x6ppler-error, tracking measurement time, pseudorange and satellite identification data. All of the tracking information must be obtained simultaneously for each satellite in view of the ranging-receiver. Furthermore, if the tracking information, which includes an accurate estimate of the time that the tracking information was measured, is transmitted (via the wireless link) to a remote site or to a wireless device, where current ephemeris data has been obtained by conventional GPS receiving equipment operating with a clear view of the sky, then, when processed with the current ephemeris data, an accurate geolocation of the wireless device can be computed.
For the purposes of this application, the term xe2x80x9cDxc3x6ppler-errorxe2x80x9d is defined as the combination of Dxc3x6ppler-shift and frequency errors in the receiver due to local oscillator instability, aging, thermal effects and manufacturing tolerances. xe2x80x9cDxc3x6ppler-shiftxe2x80x9d is the frequency shift created by the relative motion between the satellite transmitter and the ranging receiver in the device to be located. Furthermore, frequency-offset or frequency-offsets or offset-frequency or offset-frequencies may be used as appropriate in the following descriptions synonymously with Dxc3x6ppler-error or Dxc3x6ppler-errors, unless specifically modified by alternative adjectives.
Because this signal-averaging process is much longer in duration than processes used by conventional GPS receivers (estimated to be hundreds to thousands of times longer), there is potential for significant improvements in signal detection sensitivity. However, to accumulate the satellite""s tracking signal efficiently over such longer intervals, it is necessary to precisely compensate for: (i) the Dxc3x6ppler-shift induced errors (both time and frequency errors) in the signals by the relative motion between the receiver and the satellites broadcasting the signals; and (ii) frequency errors in the receiver""s timing source due to factors such as thermal effects, manufacturing tolerances and component aging.
The term xe2x80x9cranging-receiverxe2x80x9d is used herein to distinguish between: (i) the partial-function nature of a xe2x80x9cranging-receiverxe2x80x9d, which is only capable of identifying the GPS satellites and measuring pseudorange; and (ii) the full-function nature of a conventional, autonomous GPS receiver, which is capable of identifying the GPS satellites, measuring pseudorange, receiving and decoding GPS satellite ephemeris data, and making a geolocation estimate.
One attempted method for providing Dxc3x6ppler-shift information to a ranging-receiver operates on the presumption that a special xe2x80x9creference GPS receiverxe2x80x9d nearby the ranging-receiver would be able to accurately estimate and relay to the ranging-receiver the Dxc3x6ppler-shift and timing information for each satellite in order to provide the necessary Dxc3x6ppler-shift correction information, and resolve the 1 millisecond time ambiguity between ranging pulses. Such a xe2x80x9creference GPS receiverxe2x80x9d would: (i) measure the Dxc3x6ppler-shift and identity of all satellites in view at the remote location of the wireless device from a point also having a similar view of the satellites; and (ii) relay both the satellite identifications (xe2x80x9csatellite IDsxe2x80x9d) and Dxc3x6ppler-shifts and, of necessity, time-base standardizing information, via the wireless link to the GPS rangingreceiver. It is also necessary to keep track of the clock time for each satellite-in-view to allow resolution of the one-millisecond ambiguity. However, this method creates several difficulties and complexities.
For example, in order to cover a wide area, either: (i) a large number of such reference GPS receivers is needed, each such receiver providing coverage for a local radius of approximately a hundred miles (limited by the 1 millisecond ambiguity or approximately half the speed of light multiplied by the 1 millisecond interval between ranging pulses); or (ii) an information server must be provided, capable of interpolating the Dxc3x6ppler-shift information from a smaller number of reference receivers to values applicable to the location nearby to the ranging-receiver of the wireless device.
Furthermore, because Dxc3x6ppler is different at different points of measurement, the Dxc3x6ppler-shift uncertainty between two stationary ranging-receivers a hundred miles apart would be different by as much as 50 Hz, which may further limit the efficiency of the signal averaging process necessary to recover the tracking information from the weakened signals. These xe2x80x9creference GPS receiversxe2x80x9d are specialized GPS receivers capable of very precisely measuring and reporting the Dxc3x6ppler-shift of the satellite signals against a xe2x80x9cstandardizedxe2x80x9d time-base with a precision suitable for longer integration times, and capable of keeping track of the clock time for each satellite-in-view to resolve the timing ambiguity required for calculating the rangingreceiver""s geolocation. The ranging-receiver can then use the reference-receiver""s Dxc3x6ppler-shift information to compensate for ranging signals errors introduced by the Dxc3x6ppler-shift before it begins estimating the pseudorange measurement.
However, in order to estimate which reference GPS receiver is xe2x80x9cnearby enoughxe2x80x9d to the wireless device to send it sufficiently accurate Dxc3x6ppler-shift information, the system sending the Dxc3x6ppler-shift information must have a xe2x80x9ccrudexe2x80x9d estimate (i.e., within approximately one hundred miles) of the location of the wireless device. This may be difficult in some cases and near impossible in others, and certainly restricts the use of the method to systems that already have significant integration into the wireless network infrastructure (e.g., the wireless device must be sufficiently integrated in to the wireless network to report the location of the cell-site that is handling the wireless link with the wireless device).
Furthermore, the measurement of frequency implies the use of an agreed standard for the unit of time at the point of measurement and the point of use. This is almost always a difficult implication to deal with when high accuracy is required, since xe2x80x9ccommon timexe2x80x9d must be transferred from the point of measurement to the point of application or vice versa with the necessary precision and accuracy.
So, even when sufficiently accurate Dxc3x6ppler-shift information is sent to the wireless device, the sampling time-base in the wireless device must also be synchronized (i.e., calibrated or re-scaled) with the xe2x80x9cstandard time-basexe2x80x9d used to obtain the Dxc3x6ppler-shift information, or else the use of the Dxc3x6ppler-shift information will be distorted by the relative errors between the two time-bases (i.e., the xe2x80x9cstandard time-basexe2x80x9d of the reference GPS receiver and the time-base of the wireless device""s ranging-receiver). This time-base calibration or synchronization (or timescaling) normally requires xe2x80x9ctightxe2x80x9d integration of the sampling time-base (in the wireless device) with the time-base of the wireless carrier system, and/or with the time-base of the reference GPS monitoring system, to maintain the required level of accuracy and precision.
What is proposed by the applicant is a unique approach that will quickly and autonomously obtain precise Dxc3x6ppler-error correction and an alternate method of providing resolution of the timing ambiguity that does not rely on either: (i) tightly integrating with the carrier""s network; or (ii) a support network (or equivalent) of nearby reference GPS receivers to enable the ranging-receiver to provide sufficient tracking information for its geolocation to be calculated.
The present invention teaches a method whereby both the Dxc3x6ppler-error and the respective satellite""s identification can quickly (thereby minimizing the power consumption required for determining a geolocation) and precisely be determined at the wireless device, using only the signal sampling time-base in the ranging-receiver, and in circumstances where the signal""s strength has been weakened by obstructions in the signal path, thus making it possible for the wireless device to autonomously and quickly (i.e., in a matter of seconds) obtain reliable tracking information without satellite ID and Dxc3x6ppler-shift information assistance from a remote site.
The invention allows the wireless device, when connected with its carrier network, to transmit the estimated pseudorange, the time-of-day (xe2x80x9cTODxe2x80x9d) of the pseudorange measurement, satellite ID and Dxc3x6ppler-error information for each of the satellite signals being received (collectively the xe2x80x9ctracking informationxe2x80x9d) from any unknown location and anytime after making the ranging measurement (i.e. not limited to first knowing the location of points xe2x80x9cnearbyxe2x80x9d the wireless device and not requiring any high precision synchronization with any particular network""s timing system). The pseudorange, the TOD of the pseudorange measurement and the satellite HD information can be processed with the ephemeris data (collectively the xe2x80x9cgeolocating dataxe2x80x9d) of all the GPS satellites (which can be obtained, if necessary, by any set of conventional GPS receivers able to obtain GPS data from all satellites, located anywhere in the world) to compute the location of the wireless device. The Dxc3x6ppler-shift information allows the pseudorange ambiguity to be resolved by Dxc3x6ppler navigation means.
The only relative timing information required between the wireless device and the facility processing the pseudorange and satellite ID information into a geolocation, is the TOD at which the pseudorange data was measured. Depending on the specific application, most commercial applications require only modest time resolution between one to ten milliseconds.